Machine learning and AI service is an essential component of FGL’s portfolio. We have been constantly applying these techniques to tackle long-standing G&G challenges during E&P cycle, such as modeling unknown petrophysical properties through generalization of core data or predicting depositional facies using well logs, biostratigraphy, chemostratigraphy and core data as inputs.
Future Geoscience Ltd has facilitated two main visual platform to study and interpret integration data, Spotfire and I2G.
I2G cloud-based machine learning toolkit including both regression and classification enables us to employ the most current state-of-the-art machine learning methods from neural network to tree-based methods to build models for predicting wellbore data. A successful machine learning model build starts with data correction and analysis to select suitable inputs, followed by model training, model verification and finally prediction.
Depofacies Predictor is an AI tool commercially launched on i2G in 2019. A new hybrid approach which combines machine learning and an expert system generates data-driven, reproducible, and consistent results for facies prediction across an entire field, which significantly reduces uncertainties when building a conceptual model derived from facies distribution.
Apart from this, Future Geoscience Ltd is working on exploiting commercial value of the data:
- Mine model and add value to the Chemostratigraphy and biostratigraphy data library for multi-client products
- Develop refined interpretive workflows to de-risk wellsite operations and to refine completion optimisation to enhance production
- Develop interpretive workflows to reduce invasive tools in environmental impact studies
- Establish computational hub in Welshpool
- Develop AI workflows to model geological and environmental applications
- Development with Vietnam partners